I have data for two groups. The 2 groups are matched on a number of variables, but differ in the scores on an alcohol measure (one group is high, and one is low). However the groups have been created so that they capture the extremes of the alcohol measure (all very low in one group, all very high in another) therefore the alcohol measure that separates them is not strictly continuous.
I then have data for the groups on task performance in 3 different conditions: reward, neutral, and punishment. I want to compare the groups for their task performance in each of these conditions, and additionally see if each group responds differently across the conditions.
I have conducted t-tests investigating if the means are different for each group, in each different condition. The groups significantly differ in one of the conditions.
Question 1: Would it be valid to do a RM-ANOVA treating each group as independent of each other, thus using a split file in SPSS and comparing scores across the three conditions within each group?
Question 2: I am considering a MANOVA, using task performance in the 3 conditions as dependent variables and group as the independent variable. If I were to do this, would it replace the t-tests or can it be used as additional examination (whereby I would have data using a univariate approach in the t-tests, and then using a multivariate approach using the MANOVA).